Fast 3D scene object detection and real size estimation using Microsoft Kinect sensor

This paper presents an efficient and fast system for object detection in a 3D scene using the capabilities of Microsoft Kinect sensor in depth map generation. Besides, the proposed method introduces a real size estimation of the detected objects. Successful 3D scene's object detection and real size calculation are crucial features in computer vision to the goal of making machines that see objects like humans do. In our system we employ effective depth map processing techniques, along with edge detection, connected components detection and filtering approaches, in order to design a complete algorithm for efficient object detection and real size calculation, even in complex scenes with many objects. Experimental results on three different 3D scenes are presented, showing the efficiency of the proposed design.